NewYorkDave's Passive Equalizer

[MasonAtom]

I've been playing around with this a bit in LTSpice. It seems as though you can drive this from a 12AU7 cathode follower, with about 10k on the cathode. Higher resistances bring down the amount of boost you achieve (I started at 47k and only got about +5.5 dB for max).

Because the output impedance of the 12AU7 cathode follower is probably still a tad higher than Dave's recommended 100 ohms, it looks like you have to scale the total resistance of the bank if you want to maintain +10 dB boost (in this case, about 1.8k total series resistance does the trick). The frequency response and q looks really good. Here is an sim with 275Hz, 1.1k, and 5.5k based on Dave's inductor/cap choices for related frequencies. If I build this, I'll probably go for 4 or 5 frequencies per band.

I'm going to have to rescale the resistors to maintain even 2dB boosts per rotation, but that should be easy enough.

Mason

 

[MasonAtom]

Here's (a part of) the circuit.

Right now the filter section is set up in a mic preamp, following a single tube gain stage (~ 36dB gain) AC coupled to the 12AU7 cathode follower. Then Dave's filter section, followed by a 12AX7 make-up gain stage (actually I'll be using a 12DW7 which has one section of each of those triodes).

The only thing I'm wondering about right now is that in order for this to work, you need a low input impedance on the make-up gain stage. It seems to sim well with a 510R grid leak on the 12AX7, but I've never used such a low grid leak on a 12AX7 and I'm wondering if there would be unexpected complications due to this?

 

[PRR]

Your simulation should have _3_ pots/switches, set/swept independently.

R7 500 ohms is necessary and fine.

The "wrongness" is that if you can/must drive 500 ohms, you could instead replace U5 with a 500:50K transformer and get nearly as much gain with nearly as low output impedance. However the tube is cheaper than a good transformer, and offers a pinch more gain and lower output impedance.

Alternatively you could raise all impedances, but the choke values get awkward/costly.

Your cathode caps look wrong. 12AU7 with 10 ohms bias is gonna overload bad. (In .TRAN simulation you should check your model emulates a grid diode, and you should put >>10K in series with grid so you will see when the grid is gone sucky.) OTOH 3K under 12AX7 with 47K load seems large.

For quick-workup or sanity chack: Cathode bias resistors should be on the order of Rp/Mu. Your 12AU7 stage has 10K load, 10K/20= 500 ohms bias. 12AX7 stage 47K/100= 500 ohms (coincidence). This may be 330 ohms or 1K, but 10 and 3K are a bit far-out.

 

[MasonAtom]

Your simulation should have _3_ pots/switches, set/swept independently.

yes, of course. Everything presented here was the quick and simple mock-up to get the sim started. I've already begun building in more complexity as you mention.

R7 500 ohms is necessary and fine.

The "wrongness" is that if you can/must drive 500 ohms, you could instead replace U5 with a 500:50K transformer and get nearly as much gain with nearly as low output impedance. However the tube is cheaper than a good transformer, and offers a pinch more gain and lower output impedance.

yeah, I had thought about running the output from the filter section into a transformer, but I have a second half of a tube available (probably going to use 12DW7 - the 12AU7 section for the CF and the 12AX7 section as make-up gain).

Your cathode caps look wrong. 12AU7 with 10 ohms bias is gonna overload bad. (In .TRAN simulation you should check your model emulates a grid diode, and you should put >>10K in series with grid so you will see when the grid is gone sucky.) OTOH 3K under 12AX7 with 47K load seems large.

For quick-workup or sanity chack: Cathode bias resistors should be on the order of Rp/Mu. Your 12AU7 stage has 10K load, 10K/20= 500 ohms bias. 12AX7 stage 47K/100= 500 ohms (coincidence). This may be 330 ohms or 1K, but 10 and 3K are a bit far-out.

yikes, I have no idea how the 10 ohms got in there on the 12AU7. It was originally 470R.

The values on the 12AX7 are actually left over from my first sim where I had a 12AU7 there. I simply switched the .inc file and definition to 12AX7 in LTSPice and left it for the first run, knowing that I'd have to come back and rebias it. I'll likely center bias the stage with a 75k or 82k plate load.

Mason

 

[audiophreak]

Glad to finally have the opportunity to thank you for all the incredible designs you've posted here.

  I have just finished a P2P Quad of your One Bottles and could'nt be happier with them    ..  (wil post  pix soon in that thread )  .. and have been slowly accumulating parts for a Dual Mila ....  and I will definitely be building two channels of this EQ,

  Thank you so much for your insight and contributions

ps .. have you seen this thread ?  http://www.groupdiy.com/index.php?topic=45687.0

 

[Jeremy.Starseed]

This is great, and I happen to have a matched pair of 6CG7s to spare, since two weeks.

Good timing.

I would like to try this design out, if time and resources allow. Plenty of parts,
except transformers, laying around.

Thanks Dave, for providing the spark.

 

[boji]

What a wonderful return Post, NYD!  You have been missed by many people here.
I have to tell you, your summing schema was my introduction to DIY. It lead to my first build, a 16:2 passive bus I use all the time at mix-down.

Thanks again for your contributions, past and present!

 

[jwhmca]

In this design, is it true that adding series resistance (like Dave suggested) to broaden the Q, lowers the boosting ability of that Z?

 

[thakava]

Great thread!

If I wanted to build a EQ like this, but with just 100Hz (+/- 10db´s) and 5K (+/- 10db´s), how would I do that?
Also, would love to know what parts that are needed...
Am newbie when it comes to building EQ´s.

Thanks!

Tom

 

[cstella]

Dave great to see you back! You generously shared this circuit with me some years ago. Funny I just got interested in building it again and then I found this thread!

I managed to get the inductors from Wilco back then as well.

Had anyone found a good input transformer for this project that meets all the criteria?

Thanks! Charles

 

[buildafriend]

I'm happy you popped in one more time, you seem to be becoming a bit of a legend around here.

 

[pjproby]

Would this Sowter work as the input transformer:

http://www.sowter.co.uk/specs/4383.htm

I can't find all the relevent specs for this on the site...

Thank you very much!

 

[5v333]

hi!

i have found this eq, with the tube makeup stage, a very interesting thing to build for tracking and mastering.

since i lack alot of skills about inductors and tube design i have some questions and ideas i hope someone could help me with.



first is the pushpull output stage in the tube amp. im looking for output transformers that will suit but i dont know what the transformers dc current should be rated at...


another thing is the three largest inductors. since it is hard to get and expensive to buy alot of them custom made, would it work (and be cheaper?) to have  multitapped inductors made that would represent the 820mh, 330mh and the 150mh inductors togehter and wire them up to the switches together with the capacitors. i know the 150mh is for the mids while the others are for the bass but it is tied to the same line as the others so it seems like that would work.

looking forward some help with this cool project!

 

[gyraf]

You can have the 820mH and the 330mH on the same inductor - but NOT the 150mH, as this is used simultaneously with either 330 or 820 but in a different circuit..

Find a decently-high-AL core with good room for wire - e.g. the P30-type core in N30 material, the ungapped version. It's AL-value is 11500. And it's a relatively large core, which we want for low-frequency work in order not to saturate all too soon.

Remember to get a bobbin for winding the coil, and a mounting set with clips and solder pins to hold it together.

Then go to http://powermagnetics.co.uk/calculator - and put in 330mH and 11500AL, click "Calculate number of turns", and the result is 169.4 - so 170turns will give you 330mH. Now try for 820mH - gives 267 turns.

So you'll want to wind a coil of start, then tap at 170turns, then 97 turns more, ending at 267turns total.

You can estimate how heavy wire you can put on the bobbin at 267 turns, or you can just use whatever you have as 267turns won't give you excess resistance anyway with any decent size copper.

Note that you won't need any winding machine for 267 turns, and as we don't depend on filling the bobbin to 100%, it does not have to look nice - just make sure to use the entire width of the bobbin somewhat evenly.

Mount in the core (avoiding any dirt between core halves), add clips and hardware, solder wire to pins - then measure inductance while carefully "grinding" the two core halves onto each other (yes, there's a bit of slack for doing this even when mounted) to precisely align the cores and get gap down to zero. You can see the inductance rising while doing this.

Remember that core AL tolerance is +30/-20% - if you want precision matched parts you'll need to measure individual core AL before winding, give it e.g. 100 turns, mount, grind, measure, then go to above link and calculate actual AL, recalculate turn numbers...

Jakob E.

 

[5v333]

hej jakob!

thanks for your informattive and inspiring answer!!!
i think that it would be fun to roll my own inductors actually. its just that there are so many other things on my "to do list".

as i see it, all the inductors are tied to the same line. that line could be connected between the 150mh and the 330mh winding of a multitapped inductor. i have attached a drawing of how i see it.

but perhaps this is not working by the law of physics..?


anybody has a number on the current rate of the tube amp?

thanks again!!!

 

[gyraf]

No, the 150mH is series-resonant with a different set of capacitors (i.e. in use at the same time) - would give serious and strange interaction having them on the same core. You can only use one tap of the inductor at any one time.

The 150mH and the 33?mH could be on a common core though..

Jakob E.

 

[5v333]

alright then, i leave that thought and start look at cores and wire for a while.

looking at a p-core and its bobbin at mouser B65701W0000R030, it seems that wire needs to be wired in layers rather than one layer(or to use ultra thin wire), to get all the 267turns. is there some rules here or other things to consider if i would to role some inductors of my own?

 

[gyraf]

..no need for being cautious about layer stacking - you just wind until you have your count (or bobbin is full, which indicates that you need to use thinner wire)

Jakob E.

 

[5v333]

ive been looking at cores, bobins, wire, diagrams and some general googling on the subject for a couple of days now.
i have gotten some clues what to aim for/try out and what material i can get hold of.

am i right when thinking that dcr is just the wire resistance and can be precalculated after how much copper will be wound and the resistance of the specific thickness of the wire.

or is dcr measured in a more advanced way?

if the first is right then i can calculate what i need to order.

 

[gyraf]

yup, DCR is just the copper resistance. If you get too low on DCR, just add a series resistor (or simply ignore for this circuit)

Jakob E.